What is it about?
Artificial intelligence tools such as ChatGPT are increasingly used to find information, summarize research, and answer technical questions. But can their answers be trusted in financial research? This study tests ChatGPT by asking the same capital-budgeting question on three consecutive days. The results reveal two serious problems: the AI fabricated many academic references that did not exist, and its financial reasoning changed from one day to another. The study shows that convincing and professionally written AI responses may contain errors that are difficult to detect without independent verification.
Featured Image
Why is it important?
Researchers, students, and financial professionals are increasingly relying on AI to locate academic studies and analyze financial problems. This research demonstrates that AI can provide answers that appear authoritative while containing fabricated sources or inconsistent reasoning. In the experiment, approximately half of the academic references provided by ChatGPT could not be verified, and its recommendations about discount rates reversed across days. These findings highlight the importance of verifying AI-generated references, calculations, and conclusions rather than assuming that a confident and well-written response is accurate.
Perspectives
I conducted this study after encountering a practical problem that researchers increasingly face: AI can produce an answer that looks highly credible even when some of the supporting references do not exist. I wanted to examine this issue in a simple and replicable way using a financial research question where both the references and the underlying reasoning could be evaluated. What I found particularly concerning was not only the fabrication of references, but also the inconsistency of the answers when exactly the same question was asked on different days. My hope is that this study encourages researchers, students, and practitioners to use AI as a useful research tool while maintaining independent judgment and carefully verifying its output.
Dr. Hongbok Lee
Western Illinois University
Read the Original
This page is a summary of: AI Hallucination in Financial Research, International Journal of Economics and Financial Issues, September 2026, EconJournals,
DOI: 10.32479/ijefi.24171.
You can read the full text:
Contributors
The following have contributed to this page
